Paper detail

Instability Mechanism for STT-MRAM switching

To optimize the design of STT-MRAM (spin-transfer torque magnetic random access memory), it is necessary to be able to predict switching (error) rates. For small elements, this can be done using a single-macrospin theory since the element will switch quasi-uniformly. Experimental results on switching rates suggest that elements large enough to be thermally stable switch by some mechanism with a lower energy barrier. It has been suggested that this mechanism is local nucleation, but we have also previously reported a global magnetostatic instability, which is consistent with the lower experimental energy barriers. In this paper, we try to determine which of these mechanisms is most important by visualizing the switching in a "U-NU" (uniform - nonuniform) phase diagram. We find that switching trajectories follow the horizontal U axis (i.e., quasi-uniform precession) until they reach a critical amplitude, at which the magnetostatic instability grows exponentially and a domain wall forms at the center, whose motion completes the switching. We have tried unsuccessfully to induce local nucleation (a domain wall at the edge). We conclude that the dominant switching mechanism is not edge nucleation, but the magnetostatic instability.